Reasoning on infeasibility in distributed collaborative computing environment

Abstract

Collaborative computing is a new communication technology that makes it possible to extend model formulation, management, and analysis into a geographically distributed group environment. The forms of communication may vary from asynchronous hypermedia teamwork via the Internet to multipoint desktop video conferencing. The latter presents the maximal potential for integrating shared quantitative model analysis with real-time/asynchronous geographically distributed electronic meetings. Infeasibility diagnosis and reasoning on conflict resolution are known to be important parts of an evolving approach to linear programming model analysis. In the new distributed environment, several specific decision support issues related to infeasibility analysis emerge. One is the learning mechanism for capturing group knowledge on infeasibility resolution that is generated during the collaborative sessions of model analysis. The other new decision support component is a coordination protocol that is capable of linking individual activities and software transactions for the support of group reasoning on infeasibility analysis. This paper addresses these issues on the basis of committee models for infeasibility resolution and the neural network approach. Examples of modeling cases are based on experiments using a multiple criteria model for support of resource allocation in a distributed electronic meeting, and the case of model-based diagnosis that the group is involved in.